Find AI skills and prompts

Apply when the user requests integration testing, feature validation, or test plan execution.

/agent-integration-testing in Cursor’s slash menu

Finds and fixes duplicated logic, unnecessary complexity, silent errors, and inconsistent conventions in AI-written code.

/anneal in Cursor’s slash menu

Create a researched long-form article from a topic prompt via an orchestrated pipeline - research agent (first-person sources, working-definition gate).

/article-factory in Cursor’s slash menu

Runs repeated experiments and keeps changes that improve one measurable outcome.

/autoresearch in Cursor’s slash menu

Develop custom modules for [Harbor Boost](https://github.com/av/harbor/tree/main/boost), an optimizing LLM proxy.

/boost-modules in Cursor’s slash menu

Systematically explore and test any software project (CLI, API, Backend, Library, etc.) to find bugs, usability issues, and edge cases.

/bugbash in Cursor’s slash menu

Finds bugs in a defined area, verifies and fixes them, then checks the fixes against project constraints.

/bughunt in Cursor’s slash menu

Catalog-then-fix slop removal on an existing prose draft.

/catalog-deslop in Cursor’s slash menu

Gives agents a set of rules for scoping work, tracking constraints, verifying claims, and recovering from failures.

/discipline in Cursor’s slash menu

Rephrase content as a flat list of atomic facts in the style of av/facts fact sheets.

/factual-communication in Cursor’s slash menu

Runs a timed brainstorming session in which independent agents propose, critique, and compare ideas.

/ideate in Cursor’s slash menu

Reviews an existing skill, makes one focused change, and has an independent reviewer check the result.

/improve-skill in Cursor’s slash menu

Create a HyperFrames video composition from a brief/script autonomously.

/make-video in Cursor’s slash menu

Assemble a finished piece from human-written fragments via an orchestrated two-loop pipeline - an assembly loop (assembler + cold reviewer).

/mosaic-writing in Cursor’s slash menu

Build-less ESM frontends that run directly in the browser without bundlers.

/preact-buildless-frontend in Cursor’s slash menu

Downloads llama.cpp models and describes a CPU fallback for ROCm environments.

/pull-llamacpp-model in Cursor’s slash menu

Set up and run local LLMs with Harbor.

/run-llms in Cursor’s slash menu

Orchestrate AI coding agents (Claude Code, Codex, OpenCode) in isolated sandboxes using the @ai-hero/sandcastle SDK.

/sandcastle in Cursor’s slash menu

Set up and operate the Claude Code harness for large codebases.

/superclaude in Cursor’s slash menu

Carry out a task iteratively over a user-specified duration by dispatching subagents.

/timeboxed-iterating in Cursor’s slash menu

Builds a sourced research dossier by dividing a date range into year and month research passes.

/timeframe-research in Cursor’s slash menu

Deep learning framework development with tinygrad - a minimal tensor library with autograd, JIT compilation, and multi-device support.

/tinygrad in Cursor’s slash menu

Shows how to install, configure, and use Turso DB, a SQLite-compatible database engine written in Rust.

/turso-db in Cursor’s slash menu

Shows how to call installed coding agents as headless workers from a coordinating session.

/use-coding-agents in Cursor’s slash menu

Orchestrate work as a cyclic directed graph of subagent-executed nodes with transition criteria on edges, budgeted cycles, and per-node gates.

/workgraph in Cursor’s slash menu

Carry out work as an explicit event-driven state machine executed by subagents.

/workmachine in Cursor’s slash menu

Ask which skill or flow fits your situation.

/ask-matt in Claude Code after a direct install

Pursue a long-running goal in a single session by co-ordinating subagents.

/chief-of-staff in Claude Code after a direct install

Hand the current conversation off to a fresh background agent that picks up the work immediately.

/claude-handoff in Claude Code after a direct install

Reviews changes against project standards and the original specification in separate passes.

/code-review in Claude Code after a direct install

Helps choose module boundaries and interfaces that make a codebase easier to test and navigate.

/codebase-design in Claude Code after a direct install

Diagnosis loop for hard bugs and performance regressions.

/diagnosing-bugs in Claude Code after a direct install

Defines domain terms and relationships for glossaries, architecture decisions, and code.

/domain-modeling in Claude Code after a direct install

Set up Claude Code hooks to block dangerous git commands (push, reset --hard, clean, branch -D, etc.) before they execute.

/git-guardrails-claude-code in Claude Code after a direct install

A relentless interview to sharpen a plan or design.

/grill-me in Claude Code after a direct install

A relentless interview to sharpen a plan or design, which also creates docs (ADR's and glossary) as we go.

/grill-with-docs in Claude Code after a direct install

Grill the user relentlessly about a plan, decision, or idea.

/grilling in Claude Code after a direct install

Compact the current conversation into a handoff document for another agent to pick up.

/handoff in Claude Code after a direct install

Build a piece of work based on a spec or set of tickets.

/implement in Claude Code after a direct install

Build the result of /to-spec and /to-tickets in code.

/implement-spec in Claude Code after a direct install

Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.

/improve-codebase-architecture in Claude Code after a direct install

Grill me about specs for the workflows I want to build, within this workspace.

/loop-me in Claude Code after a direct install

Migrate test files from `as` type assertions to @total-typescript/shoehorn.

/migrate-to-shoehorn in Claude Code after a direct install

Drafts a pull request body that explains the change, evidence, and risk.

/pr in Claude Code after a direct install

Builds contrasting interface concepts so teams can compare options before committing to a direction.

/prototype in Claude Code after a direct install

Investigate a question against high-trust primary sources and capture the findings as a Markdown file in the repo.

/research in Claude Code after a direct install

Reviews a completed coding session to capture lessons and improve future work.

/retro in Claude Code after a direct install

Develop exercise directory structures with sections, problems, solutions, and explainers that pass linting.

/scaffold-exercises in Claude Code after a direct install

Set up this repo for the engineering skills: set up its issue tracker, triage label vocabulary, and domain doc layout.

/setup-matt-pocock-skills in Claude Code after a direct install

Set up Husky pre-commit hooks with lint-staged (Prettier), type checking, and tests in the current repo.

/setup-pre-commit in Claude Code after a direct install

Wire dependency-cruiser into a TypeScript repo so each package is a deep module, with implementation hidden in subfolders and reachable.

/setup-ts-deep-modules in Claude Code after a direct install

Builds features through a test-first cycle that checks behavior before and after implementation.

/tdd in Claude Code after a direct install

Teach the user a new skill or concept, within this workspace.

/teach in Claude Code after a direct install

Convert a decision you can't fully answer into a questionnaire for someone else to fill in.

/to-questionnaire in Claude Code after a direct install

Convert the current conversation into a spec and publish it to the project issue tracker: no interview, just synthesis of.

/to-spec in Claude Code after a direct install

Stress-test a plan, spec, or the current conversation into a set of tracer-bullet tickets, each declaring its blocking edges, published.

/to-tickets in Claude Code after a direct install

Move issues and external PRs through a state machine of triage roles, categorise, verify, grill if needed, and write agent-ready.

/triage in Claude Code after a direct install

Rephrases confusing messages in clearer language so the conversation can move forward.

/wait-what in Claude Code after a direct install

Shape a huge chunk of work (more than one agent session can hold) as a shared map of decision tickets.

/wayfinder in Claude Code after a direct install

Produce an interactive bash wizard that walks a human through steps only they can perform.

/wizard in Claude Code after a direct install

Writing, exploit; assemble raw material into a journey of beats, grounding each term before a beat leans on it.

/writing-beats in Claude Code after a direct install

Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.

/writing-for-agents in Claude Code after a direct install

Writing, explore: mine raw fragments, no structure yet.

/writing-fragments in Claude Code after a direct install

Writing, exploit: shape raw material into an article, paragraph by paragraph.

/writing-shape in Claude Code after a direct install

Starts creative work with a structured idea session to clarify the problem and compare possible directions.

/superpowers:brainstorming in Claude Code with the Superpowers plugin

Investigates failed Superpowers sessions, including repeated work and ignored plans.

/superpowers:diagnosing-superpowers in Claude Code with the Superpowers plugin

Apply when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies.

/superpowers:dispatching-parallel-agents in Claude Code with the Superpowers plugin

Carries out an implementation plan in the current session with checkpoints for review.

/superpowers:executing-plans in Claude Code with the Superpowers plugin

Apply when implementation is complete, all tests pass, and you need to decide how to integrate the work.

/superpowers:finishing-a-development-branch in Claude Code with the Superpowers plugin

Apply when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical.

/superpowers:receiving-code-review in Claude Code with the Superpowers plugin

Apply when completing tasks, implementing major features, or before merging to verify work meets requirements.

/superpowers:requesting-code-review in Claude Code with the Superpowers plugin

Apply when executing implementation plans with independent tasks in the current session.

/superpowers:subagent-driven-development in Claude Code with the Superpowers plugin

Apply when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

/superpowers:systematic-debugging in Claude Code with the Superpowers plugin

Apply when implementing any feature or bugfix, before writing implementation code.

/superpowers:test-driven-development in Claude Code with the Superpowers plugin

Sets up an isolated Git worktree for a feature or implementation plan.

/superpowers:using-git-worktrees in Claude Code with the Superpowers plugin

Apply when starting any conversation - establishes how to find and use skills, requiring skill invocation before ANY response including.

/superpowers:using-superpowers in Claude Code with the Superpowers plugin

Apply when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification.

/superpowers:verification-before-completion in Claude Code with the Superpowers plugin

Apply when you have a spec or requirements for a multi-step task, before touching code.

/superpowers:writing-plans in Claude Code with the Superpowers plugin

Apply when creating new skills, editing existing skills, or verifying skills work before deployment.

/superpowers:writing-skills in Claude Code with the Superpowers plugin